项目来源

俄罗斯科学基金(RSF)

项目主持人

Berdyugin Aleksandr

项目受资助机构

Tomsk State University

项目编号

24-79-00314

立项年度

2024

立项时间

未公开

研究期限

未知 / 未知

项目级别

国家级

受资助金额

未知

学科

ENGINEERING SCIENCES-Development of new structural materials and coatings

学科代码

09-09-205

基金类别

未公开

Терагерцы ; искуственный интеллект ; аддитивность ; дефектоскопия ; спектроскопия ; визуализация ; 3D печать ; нейронные сети ; Terahertz ; artificial intelligence ; additivity ; defectoscopy ; spectroscopy ; visualization ; 3D printing ; neural networks

参与者

未公开

参与机构

未公开

项目标书摘要：nnotation:The use of additive technologies makes it possible to put quality characteristics into the design of electronic equipment enclosures and provide radio transparency or shielding from electromagnetic fields of radioelectronic equipment blocks and their connections when using polymers and electrically conductive materials.In the process of production of dielectric enclosures of radio electronic equipment inevitably arise defects characteristic of the specific technology used.Their appearance leads to a decrease in both structural and consumer qualities of products.Defects can be divided into two groups:obvious and hidden.The current state of the art makes it possible to detect the former with a high probability of detection and to do it in a flow mode.For example,computer vision technologies are widely used for this purpose.Explicit defects in the cases of radio-electronic equipment,usually manifest themselves before the stage of operation and do not lead to an increase in failures of equipment.Hidden defects,on the contrary,are not easy to detect.The equipment for their diagnostics is expensive and most often based on X-ray tomography techniques,which,in turn,imposes high requirements to the safety of personnel in production.Promising is the application in this area of terahertz radiation,which has no ionizing character and has a high penetrating ability in dielectric media.These facts indicate the possibility of creating a system of terahertz diagnostics of radio electronic equipment housings,in which by spatial registration of the past(reflected)radiation intensity it is possible to obtain information about the internal structure of the product housing.An integral part of the nondestructive testing system should be an automated subsystem responsible for image processing of the investigated objects.After image processing,the subsystem should make a decision and provide information on the presence or absence of defects.Machine learning methods are widely used in the field of terahertz technology,mainly as tools for terahertz spectroscopy and image preprocessing,as well as methods for qualitative and quantitative analysis of multivariate data.Machine learning has been pervasive in many applications and has demonstrated high performance and problem-specific results beyond the capabilities of current methods available today.The aim of the present project is to develop a non-contact method for diagnosing defects in housing elements of radio-electronic equipment made of thermoplastic materials with functional additives for automatic detection and subsequent localization of inhomogeneities based on the results of analysis of the distribution of terahertz response parameters from the object using machine learning methods.It is planned to attract two performers to realize the project:Diana Andreevna Pidotova,a 1st year postgraduate student of the Radiophysical Faculty of TSU in the direction of"Radiophysics";Alexander Vyacheslavovich Perevalov,a 5th year student of the Radiophysical Faculty of TSU in the direction of"Radioelectronic Systems and Complexes".Expected results:As a result of the project it is planned to develop a method of non-destructive diagnostics of defects of housing elements of radio-electronic equipment made of thermoplastic materials with functional additives.Based on the results of analysis of the distribution of terahertz response parameters from the object under study,it is planned to implement a method of automatic detection and subsequent localization of inhomogeneities.The detection and classification of inhomogeneities will be based on the principles of machine learning based on neural networks.Currently,such studies are common in many applications and have demonstrated high performance in solving specific tasks.It is expected that the developed hardware and software solution for nondestructive terahertz defectoscopy of objects using a neural network will correspond to the world level and is promising for implementation in the technological process in the production of composite housings of radioelectronic equipment for spatial identification of local defects and their classification.

Application Abstract: Annotation:The use of additive technologies makes it possible to put quality characteristics into the design of electronic equipment enclosures and provide radio transparency or shielding from electromagnetic fields of radioelectronic equipment blocks and their connections when using polymers and electrically conductive materials.In the process of production of dielectric enclosures of radio electronic equipment inevitably arise defects characteristic of the specific technology used.Their appearance leads to a decrease in both structural and consumer qualities of products.Defects can be divided into two groups:obvious and hidden.The current state of the art makes it possible to detect the former with a high probability of detection and to do it in a flow mode.For example,computer vision technologies are widely used for this purpose.Explicit defects in the cases of radio-electronic equipment,usually manifest themselves before the stage of operation and do not lead to an increase in failures of equipment.Hidden defects,on the contrary,are not easy to detect.The equipment for their diagnostics is expensive and most often based on X-ray tomography techniques,which,in turn,imposes high requirements to the safety of personnel in production.Promising is the application in this area of terahertz radiation,which has no ionizing character and has a high penetrating ability in dielectric media.These facts indicate the possibility of creating a system of terahertz diagnostics of radio electronic equipment housings,in which by spatial registration of the past(reflected)radiation intensity it is possible to obtain information about the internal structure of the product housing.An integral part of the nondestructive testing system should be an automated subsystem responsible for image processing of the investigated objects.After image processing,the subsystem should make a decision and provide information on the presence or absence of defects.Machine learning methods are widely used in the field of terahertz technology,mainly as tools for terahertz spectroscopy and image preprocessing,as well as methods for qualitative and quantitative analysis of multivariate data.Machine learning has been pervasive in many applications and has demonstrated high performance and problem-specific results beyond the capabilities of current methods available today.The aim of the present project is to develop a non-contact method for diagnosing defects in housing elements of radio-electronic equipment made of thermoplastic materials with functional additives for automatic detection and subsequent localization of inhomogeneities based on the results of analysis of the distribution of terahertz response parameters from the object using machine learning methods.It is planned to attract two performers to realize the project:Diana Andreevna Pidotova,a 1st year postgraduate student of the Radiophysical Faculty of TSU in the direction of"Radiophysics";Alexander Vyacheslavovich Perevalov,a 5th year student of the Radiophysical Faculty of TSU in the direction of"Radioelectronic Systems and Complexes".Expected results:As a result of the project it is planned to develop a method of non-destructive diagnostics of defects of housing elements of radio-electronic equipment made of thermoplastic materials with functional additives.Based on the results of analysis of the distribution of terahertz response parameters from the object under study,it is planned to implement a method of automatic detection and subsequent localization of inhomogeneities.The detection and classification of inhomogeneities will be based on the principles of machine learning based on neural networks.Currently,such studies are common in many applications and have demonstrated high performance in solving specific tasks.It is expected that the developed hardware and software solution for nondestructive terahertz defectoscopy of objects using a neural network will correspond to the world level and is promising for implementation in the technological process in the production of composite housings of radioelectronic equipment for spatial identification of local defects and their classification.